Deep within Apple’s systems is a variety of instructions it has given to its GenAI Apple Intelligence mechanism. The screen captures of those instructions provide a peek into Apple’s efforts to influence its GenAI deployment, and also illustrate the steep challenges in controlling an algorithm that is simply trying to guess answers. 

The more explicit and contained an instruction, the easier it is for GenAI to understand and obey it. Therefore, some of the Apple instructions, such as “You prefer to use clauses instead of complete sentences”, and “Please keep your summary of the input within a 10-word limit”, should work well, AI specialists said.

But other, more interpretable commands from the Apple screen captures, such as “Do not hallucinate. Do not make up factual information,” may not be nearly as effective.

“I have not had good luck telling it not to hallucinate. It’s not clear to me that it knows when it is hallucinating and when it is not. This thing isn’t sentient,” said Michael Finley, CTO at AnswerRocket. “What does work is to ask it to reflect on its work, or to use a second prompt in a chain to check the results of the first one. Asking it to double check results is common. This has a verifiably good impact on results.”

Finley was also baffled at a comment that told the system to “only output valid JSON and nothing else.” 

“I am surprised that they told it to only use valid JSON. The model is either going to use it or not,” Finley said, adding it has no practical or meaningful way to assess validity. “The whole thing is really unsophisticated. I was surprised that this is what is at the heart.” He concluded that “it was kind of cobbled together. That is not necessarily a bad thing.” By that he meant that Apple developers were under pressure to move the software out quickly.

The instructions under scrutiny were for new GenAI capabilities being built into Apple’s Siri. The dataset Apple will be using is far larger than earlier efforts, which is why it will only be available on the latest devices with the strongest CPU horsepower as well as the most RAM.

“Apple’s models for Siri have been small until now. Using GPT — arguably some of the largest models — means new capabilities,” Finley said. “As parameter counts get bigger, models learn to do things that are more indirect. Small models can’t role-play, larger models can. Small models don’t know about deception, larger models do.”

Clyde Williamson, product security architect at Protegrity, was amused by how the existence in a public forum of the comments, which were presumably not intended to be seen by Apple customers, nicely illustrates the overall privacy/data security challenges within GenAI.

“This does highlight, though, the idea of how security in AI becomes a bit fuzzy. Anything we tell an AI, it might tell someone else,” Williamson said. “I don’t see any evidence that Apple tried to secure this prompt template, but it’s reasonable to expect that they didn’t intend for end-users to see the prompts. Unfortunately, LLMs are not good at keeping secrets.”

Another AI specialist, Rasa CTO Alan Nichol, applauded many of the comments. “It was very pragmatic and simple,” Nichol said, but added that “a model can’t know when it’s wrong.”

“These models produce plausible texts that sometimes overlap with the truth. And sometimes, by sheer accident and coincidence, it is correct,” Nichol said. “If you think about how these models are trained, they are trying to please the end-user, they are trying to think of what the user wants.”

Nichol liked many of the comments, though, noting, “The instructions to keep everything short, I always use comments like that,” because otherwise, LLMs tend to be “incredibly verbose and fluffy.”

Enterprises are becoming increasingly impressed by the robust security of Macs, and Apple is locking its platform down even more firmly with macOS Sequoia and a couple of changes to improve defenses against malware and “camfecting.” This reflects the company’s continued mission to ensure platform security by design.

Gatekeeper empowerment

The first change is the biggest. Apple’s Gatekeeper protection is designed to stop people from running unsafe applications on their Macs. When you try to install software downloaded from the Internet, you are presented with a security warning before the application will work (though it has long been possible for Mac users to bypass the protection by Control-Clicking on the application icon).

Apple has abandoned this in the latest Sequoia beta. Now, users must actively open Settings > Privacy & Security to permit their system to run such apps on a per-app basis. 

While the impact of this change is slight — you can still install and use apps obtained elsewhere — it should help prevent users from accidentally installing malware because it makes the whole process more intentional. Less-experienced users become less likely to be tricked into giving such approval by the app installation screen.

Apple recommends notarization

The real aim of the change is to prevent users who might be less tech-savvy from being tricked into bypassing Gatekeeper. In an ideal world, Apple would like all apps installed on Macs to at least notarized, the company confirms.

“If you distribute software outside of the Mac App Store, we recommend that you submit your software to be notarized,” Apple says. “The Apple notary service automatically scans your Developer ID-signed software and performs security checks. When your software is ready for distribution, it’s assigned a ticket to let Gatekeeper know it’s been notarized so customers can run it with confidence.”

This is a similar process to what Apple is trying to achieve on iOS devices in Europe. The goal is to secure the user and the platform, while also narrowing the size of the attack surface on its systems.

Camfecting and how to stop it

The second change will seem annoying to some, but does at least put Mac users in control. If you have ever installed screen recording or video conferencing software, you were probably asked to provide permission for those applications to capture your Mac screen. You likely went ahead and gave that permission and forgot about it — but that means applications you (or someone with access to your Mac) gave such permission to might be able to secretly continue recording your actions.

This improves in macOS Sequoia, which will require that you review and confirm this permission once a week. A dialog box will appear explaining the app wants to access the computer screen and audio, and giving you two choices: disable that permission, or “Continue to Allow” access.

While some might see this process as overly intrusive, it should help protect Macs against some in-person and malware-based camfecting attacks, as any application that has permission to access the camera/screen recording will be surfaced once a week. That means if an app you didn’t expect to see there appears on the list, you should take immediate steps to secure your device.

User controlled security

Seen in context, these latest security improvements mean the Mac is becoming better locked down as Apple works to make security protections you already have in place more understandable.

Take the Privacy & Security section of Settings for example: Over time, this has become an extensive, perhaps daunting, collection of options Apple has made easier to understand. In Sequoia, you can now more easily see how many apps enjoy full or partial access to the various settings and have a guide to help you manage those settings.

Again and again with its security improvements, Apple continues working to make security an intentional choice, explains what it is users are securing, and is creating device management APIs IT can use to ensure that their entire fleet remains as secure as it can possibly be — no kernel access required.

Please follow me on Mastodon, or join me in the AppleHolic’s bar & grill and Apple Discussions groups on MeWe.

Long before Taco Tuesday became part of the pop-culture vernacular, Tuesdays were synonymous with security — and for anyone in the tech world, they still are.  Patch Tuesday, as you most likely know, refers to the day each month when Microsoft releases security updates and patches for its software products — everything from Windows to Office to SQL Server, developer tools to browsers.

The practice, which happens on the second Tuesday of the month, was initiated to streamline the patch distribution process and make it easier for users and IT system administrators to manage updates.  Like tacos, Patch Tuesday is here to stay.

In a blog post celebrating the 20th anniversary of Patch Tuesday, the Microsoft Security Response Center wrote: “The concept of Patch Tuesday was conceived and implemented in 2003. Before this unified approach, our security updates were sporadic, posing significant challenges for IT professionals and organizations in deploying critical patches in a timely manner.”

Patch Tuesday will continue to be an “important part of our strategy to keep users secure,” Microsoft said, adding that it’s now an important part of the cybersecurity industry.  As a case in point, Adobe, among others, follows a similar patch up cadence.

Patch Tuesday coverage has also long been a staple of Computerworld’s commitment to provide critical information to the IT industry. That’s why we’ve gathered together this collection of recent patches, a rolling list we’ll keep it updated each month!

In case you missed a recent Patch Tuesday announcement, here are the latest six months of updates.

July: four zero-day flaws

This July’s Patch Tuesday from Microsoft addressed a significant number of vulnerabilities, including four zero-day threats.  Here’s a quick rundown: Microsoft released updates for SQL Server, with patches for Windows, Office, .NET, and Visual Studio. It also released four    critical updates for Windows, including patches for Hyper-V and MSHTML. One critical update for Office’s SharePoint platform.
More info on Microsoft Security updates for July 2024.

June: Relatively quiet on major updates

This month’s Patch Tuesday brought mostly low-risk updates with no reported zero-day vulnerabilities. Key areas addressed include changes to Secure Boot (requiring third-party driver testing), code integrity policies (needing verification for Windows Defender features), and core Windows systems (necessitating broad application testing). While there were no critical updates for Office or Exchange Server, some updates to Visual Studio require attention for developers.
More info on Microsoft Security updates for June 2024

May:  3 zero-day vulnerabilities signal ‘patch now’ alert

This month’s Patch Tuesday highlights three critical zero-day vulnerabilities affecting Windows PCs and requiring immediate patching  — that is,  identified as “patch now.” Some updates like those to Office and Edge browsers follow standard release schedules, but be aware of a critical update for SharePoint Server.  Developers need to aware o a late addition to the update cycle affecting the Azure Agent, requiring attention for Azure-based virtual macHines. Testing is crucial this month, especially for core Windows features like the Common Error Log, DNS, cryptography and routing services. More info on Microsoft Security updates for May.

April: Microsoft showers users with 149 patches

April’s Patch Tuesday was a complex one, especially for SQL-dependent applications. This hefty Patch Tuesday from Microsoft included 149 updates. While there were no zero-day vulnerabilities, key areas addressed include crypto APIs, networking and remote desktop connections. A major update to the Kerberos security system removes Windows 11 from the affected list, highlighting the importance of staying updated. For developers, 11 updates target the development platform, with 10 focused on SQL ODBC issues and 1 on .NET. While the .NET update can be added to the standard schedule, the ODBC updates require careful examination.
More info onMicrosoft Security updates for April.

March: It’s a complicated Patch Tuesday

This month’s Patch Tuesday from Microsoft was complexThere were no reported zero-day vulnerabilities, but number of updates, particularly those affecting SQL, OLE and ODBC components, underscores the importance of a thorough evaluation. Key areas of focus include file management, cryptography, networking, remote desktop connections and SQL-related functionalities. Given the interconnectedness of these systems, organizations should prioritize testing across their application portfolios to identify potential impacts. The update to the Kerberos security system is noteworthy, as it removes support for certain Windows 11 versions.
More info on Microsoft Security updates for March.

February : Outlook and Exchange Server hit with updates

February’s Patch Tuesday from Microsoft was significant with a critical combination of vulnerabilities affecting Microsoft Outlook and Exchange Server, both actively exploited. These updates required immediate patch now. While most updates are rated important and can be added to standard release schedules, pay close attention to the recently exploited Windows SmartScreen vulnerability and update Windows immediately. For developers, core .NET and Visual Studio updates require standard release schedule inclusion. A new SignalR library was introduced for real-time web functionality in ASP.NET.
More info on Microsoft Security updates for February.

Dell Technologies is reportedly laying off thousands of staffers from its sales team.

The layoffs are part of the hardware vendor’s reorganization efforts targeted at increasing its revenue from AI-related sales by putting in a new sales unit focused on AI products, according to a Bloomberg report.

“We are getting leaner,” sales executives Bill Scannell and John Byrne were quoted as saying in an internal memo to Dell employees.

 “We’re streamlining layers of management and reprioritizing where we invest,” the executives explained, adding that the company would be changing its approach towards data center sales.

In an emailed statement to Computerworld, Dell said, “Through a reorganization of our go-to-market teams and an ongoing series of actions, we are becoming a leaner company. We are combining teams and prioritizing where we invest across the company. We continually evolve our business so we’re set up to deliver the best innovation, value and service to our customers and partners.”

In March of this year, the company revealed in a filing that it was continuing to execute cost management measures, such as workforce reductions, limiting external hiring, and employee reorganizations, as it tried to navigate issues related to the ongoing macroeconomic environment.

The layoffs in March resulted in a reduction of the company’s overall headcount. After their completion, the employee count stood at approximately 120,000, declining from the 133,000 recorded in February 2023.

A repeat of last year

During the first week of February last year, the company laid off 6,650 workers, about 5% of its total workforce, due to declining PC sales and infrastructure requirements.

The company had already tried to cut costs by pausing hiring and limiting travel before taking the decision to downsize its workforce, Co-Chief Operating Officer Jeff Clarke shared in a blog post at the time.

In addition to the downsizing, Clarke said the company would introduce a slew of changes that include changing the structure of its sales team and integrating the services division of its consumer and infrastructure businesses.

“We have further opportunity to drive efficiency through department reorganizations, which has resulted in a reduction of team members across the globe,” a company spokesperson said in an email after the 2023 layoffs.

Tech industry layoffs continue

Dell Technologies is not the only company that has had to lay off staffers this year, especially in its sales team, to maintain business operations.

In January, Google, in an extension to the restructuring strategy it planned last year to maintain investor confidence and adjust to market conditions, laid off a few hundred staffers from its advertising sales team and started relying on machine learning to automate advertising.

After two years of massive layoffs at IT companies, 2024 was expected to be a year of recovery for the industry. While there are early signs of that, with global IT spending expected to increase 8% to cross $5.1 trillion in 2024 according to Gartner, jobs continue to be impacted in the sector.

So far in 2024, 387 tech companies globally have laid 126,032 employees, according to job reduction tracker website layoffs.fyi.

Those companies include the likes of Intel, Cisco, Microsoft, SAP, and Docusign.

In a landmark decision, a US District Court on Monday ruled that Google is a monopoly that used its dominance in the online search market to suppress other search engines and keep them from gaining market share.

The decision by the US District Court for the District of Columbia said Google’s dominance has gone unchallenged for well over a decade — allowing it to grow even stronger at the expense of its competitors.

“By 2020, it was nearly 90% [of the search market], and even higher on mobile devices at almost 95%” US District Judge Amit Mehta wrote in his 277-page opinion. “The second-place search engine, Microsoft’s Bing, sees roughly 6% of all search queries — 84% fewer than Google. Google has not achieved market dominance by happenstance.

“Google is a monopolist, and it has acted as one to maintain its monopoly,” Mehta added.

Google Search is parent-company Alphabet’s oldest and most profitable business.

After a lengthy trial last fall, the court found Google has spent tens of billions of dollars on exclusive contracts to secure a dominant role as the default search engine for web browsers and mobile devices. The court even used a competitive analysis performed by Google itself in 2020 to illustrate the point; that study estimated it would cost Google competitor Apple $20 billion to create a similar product that could compete with Google search.

Other rivals, such as Microsoft Bing and Duckduckgo.com, suffer the same inability to compete — and as Google’s revenue has grown, so has its ability to continue locking its competition out.

“In 2014, Google booked nearly $47 billion in advertising revenue. By 2021, that number had increased more than three-fold to over $146 billion. Bing, by comparison, generated only a fraction of that amount — less than $12 billion in 2022,” Mehta wrote.

While Mehta didn’t offer a solution to the monopoly Google has built, the court’s decision will likely affect other lawsuits against the company and change the way its search engine works in the future. If it prevails through the appeals process, the ruling is also likely open the door to Google’s competition to grow their own business and search engine market base.

Google’s court woes are far from over with the Monday ruling. A second lawsuit by the Department of Justice brought by the Biden administration will challenge the company’s advertising technology business.

The DOJ complaint has alleged that Google monopolizes advertising and uses technologies to eliminate or severely diminish any threat to its advertising dominance. That case is expected to head to trial in early September.

After the ruling was made public, Attorney General Merrick Garland weighed in with a statement: “This victory against Google is an historic win for the American people,” said Garland. “No company — no matter how large or influential — is above the law. The Justice Department will continue to vigorously enforce our antitrust laws.”

“This landmark decision holds Google accountable,” said Assistant Attorney General Jonathan Kanter. “It paves the path for innovation for generations to come and protects access to information for all Americans.”

Google did not immediately have comment on the landmark ruling.

Google’s dominance in the search arena has given rise to two major antitrust lawsuits from the U.S. government, which allege that the company has manipulated the market to maintain that dominance, to the exclusion of its competitors and the detriment of the public at large.

The first lawsuit, targeting Google’s search business, kicked off in mid-September 2023, and drew to a close in May 2024 with the delivery of closing arguments; a second trial against the tech giant, focusing on advertising, is scheduled for later this year.

The cases heavily echo the turn-of-the-century Microsoft antitrust case in several respects, not least of which is the fact that Google faces the possibility of being broken up by regulators if it is unsuccessful in its legal battles.

Here’s our condensed timeline of the two lawsuits, and their progress through the court system.

Aug. 5, 2024: In a major defeat for Google, Judge Amit Mehta ruled that the company had engaged in anticompetitive behavior in an effort to protect its search business. In the 277-page decision, Mehta was blunt: “After having carefully considered and weighed the witness testimony and evidence, the court reaches the following conclusion: Google is a monopolist, and it has acted as one to maintain its monopoly. It has violated Section 2 of the Sherman Act.” Attorney General Merrick Garland, in a statement from the Department of Justice, said: “This victory against Google is an historic win for the American people. No company — no matter how large or influential — is above the law.” Mehta’s ruling did not include remedies for the anticompetitive behavior; those will be decided later.

May 3, 2024: Over two days of closing arguments, the Department of Justice revisited its case for Google having a monopoly on search advertising, and Judge Mehta quizzed both parties about whether other platforms could be viewed as substitutes for Google’s search advertising business. He hasn’t said how long he expects to take to reach a decision, but if he rules against Google, a second hearing will take place to decide on any remedies.

November 16, 2023: The evidentiary phase of the trial finishes, as Judge Mehta issues instructions for post-trial submissions. Despite considerable amounts of redaction and closed-door testimony, the case revealed some unprecedented details about the relationships between the largest tech companies in the world, including the fact that Apple apparently keeps 36% of the search revenue from Google searches in Safari, and Apple once considered buying Microsoft’s Bing search engine as leverage against Google. Judge Mehta has scheduled closing arguments in the case for May 1, 2024.

October 31, 2023: Google CEO Sundai Pichai takes the stand, for long-awaited testimony about the relationship between his company and Apple. He gave some details about Google’s negotiations with Apple over a contract that made Google the default search engine on Apple’s iPhones, iPads, and Macs. Google has paid billions for the privilege of being the default search on Apple products, and the relationship is a key part of the case – which was underlined by the Justice Department’s cross-examination of Pichai, during which he admitted that default search status is a major driver of market share.

October 18, 2023: Google begins its defense, calling Paul Nayak, a vice president of search, to the stand as its first witness. Nayak downplays the importance of scale in his testimony, stressing that machine intelligence, compute infrastructure, and a team of 16,000 staff that checks on search results are crucial to maintaining quality of service. DOJ witnesses including DuckDuckGo CEO Gabriel Weinberg and Microsoft CEO Satya Nadella had testified that Google keeps an edge over competitors via an ever-increasing trove of data — the result of its default search engine status, maintained through exclusive contracts and billions of dollars in payments to Apple, Samsung and other companies. This data gives Google an advantage in refining search engine results, they said. 

October 3, 2023: As a witness for the prosecution in the Google antitrust trial, Microsoft CEO Satya Nadella warns that Google’s monopoly profits could lock in publishers as AI-enabled search arrives. Nadella argued that it’s almost impossible to compete with Google, given the search leader’s massive competitive edge in collecting and analyzing user data. He also warned that Google, with its vast profits and lock on the search market, stands poised to extend its monopoly power in a new era where artificial intelligence technologies will turbocharge the search business.

September 26, 2023: Apple’s Eddy Cue testifies behind closed doors in the Google search case, as critics slam presiding Judge Amit Mehta’s decision to hold much of the trial’s testimony from witnesses secret, allow documents to be heavily redacted, and block some documents from public view — mainly at the insistence of Google, but also at the request of other companies, including Apple. By the end of Cue’s testimony — and after a wek of wrangling by all parties — Judge Mehta rules that documents used during the trial can be published online at the end of each day, but still allows time Google and third parties to object to exhibits being shown publicly before the DOJ presents them in court. 

September 21, 2023: Judge Mehta rules that public access to court exhibits, which have been mostly internal Google documents thus far, should be removed, after Google challenged the Justice Department’s regular publication of them. The company said that it was concerned for its employees’ privacy.

September 12, 2023: The default search trial begins with opening statements, and the government begins its case.

August 2023: Judge Mehta grants partial summary judgment for Google in the search case, saying that the government had failed to raise a genuine dispute of material fact on antitrust charges relating to contracts around the use of the Android operating system, as well as Google Assistant and IoT devices. The claims relating to Google’s exclusive “default search” contracts, however, are allowed to proceed to trial.

July/August 2023: Google and the plaintiffs in the search case argue various motions in limine, designed to control what evidence should be included or excluded in the actual trial. Discovery and motion practice over evidence continues in the advertising case.

June 2023: Judge Mehta schedules a trial date of September 12, 2023 for the search case.

April 2023: Judge Leonie M. Brinkema denies Google’s motion to dismiss in the advertising case.

March 2023: Google’s motion to transfer the advertising case to New York is denied by Judge Brinkema, who orders the parties to propose discovery schedules within two weeks of the order. Two weeks later, Google moves to dismiss the case for failure to state a claim, arguing that the plaintiffs have simply produced legal conclusions, and not specific facts, that could support their claims. Judge Brinkema schedules pre-trial conferences for January 2024.

February 2023: The plaintiffs in the default search case case move for sanctions against Google, accusing it of spoliation, which refers to the destruction, alteration or failure to preserve relevant evidence in a case. Elsewhere, in the advertising case, Google moves to transfer the case from the Eastern District of Virginia to the Southern District of New York, which is seen as an attempt to consolidate the case with related digital advertising antitrust litigation.

January 2023: A second antitrust action, this one filed by eight states and the DoJ, is filed in federal district court in eastern Virginia. The plaintiffs, who call for Google’s advertising business to be split up, accuse Google of manipulating its dominant position in the online advertising world to squeeze out rivals and control both the supply and demand side of the advertising market. Google, according to the complaint, thwarted fair competition by manipulating fees, punished advertisers for using alternative platforms and ad exchanges, and engaged in a host of further anti-competitive behavior in the interest of monopolizing the marketplace.

December 2022: Google moves for summary judgment against the separate Colorado case and the larger, DoJ-led case. A summary judgement motion is essentially a request by one of the parties in a lawsuit that the judge rule in their favor and end the case, arguing that, based on the undisputed facts, they are entitled to win the case as a matter of law.

May 2022: A deadline of June 17 is set for the production of all discovery materials. Further documents – for example, those whose is existence is first disclosed in late in the discovery window – can be produced until June 30.

May 2022: Judge Mehta denies a government motion to sanction Google for inaccurately classifying documents as attorney-client privileged. The plaintiffs had argued that emails on which Google’s lawyers were listed as recipients or CCed, but that the lawyers never responded to, constituted a misuse of the attorney-client privilege rules.

December 2021: Judge Mehta conditionally splits Colorado’s claims from the case at large, ordering that separate trials on that state’s issues of liability and remedies will be “more convenient for the Court and the Parties, and will expedite and economize this litigation.”

August-October 2021: Discovery-related motions and orders continue, as Yelp and Samsung join the fray. (Those companies, like Microsoft and Apple, are relevant to the case even if they aren’t parties themselves, as their internal records are potentially relevant to Google’s liability.)

June/July 2021: The discovery process continues, and the U.S. and Google both file several documents with the court under seal. (Microsoft files two sealed documents, as well, in response to Google’s subpoenas for company records, and Apple becomes involved after the government requests access to some of its internal information.)

March 2021: Meetings between Google and the various governmental plaintiffs continue, with periodic status reports on the discovery process.

January 2021: Google files a response to the complaint, admitting to many of the facts alleged by the Justice Department and associated attorneys general, but categorically denying the substance of the government’s claims of illegality. Further responses to separate but related claims, generally to specific state attorneys general, follow in the subsequent weeks and months.

December 2020: Judge Amit Mehta approves the joinder of Michigan, Wisconsin and California to the suit.

October 2020: The Department of Justice, along with the attorneys general of 11 states, sues Google in DC federal district court for unlawfully maintaining a monopoly, in violation of Section 2 of the Sherman Act. The case centers on Google’s use of exclusive contracts that mandate its use as the default search engine in a host of different hardware and software applications, with the government alleging that this represents an artificial constraint on any possible competition for the search giant.

A probe by the US Department of Justice (DOJ) into Nvidia’s selling practices will likely go nowhere, Scott Bickley, advisory practice lead at Info-Tech Research Group, predicted Friday.

On Thursday, Reuters first reported findings of an earlier published report that “DOJ investigators are looking at whether Nvidia pressured cloud providers to buy multiple products,” and in a follow-up article stated that 10 progressive groups, and Democratic Senator Elizabeth Warren, have also asked the DOJ to investigate the company, citing its “dominance of the market for the chips driving the artificial-intelligence boom.”

In a joint letter to Assistant Attorney General Jonathan Kanter, sent on Tuesday, the groups welcomed news of plans to open an antitrust investigation into Nvidia.

They stated that the company “has made it clear that it intends to ride the AI wave as long and as far as it can, and its astonishing dominance in GPU accelerator chips — it now holds an 80% overall global market share in GPU chips and a 98% share in the data center market — puts it in a position to crowd out competitors and set global pricing and the terms of trade. Given the integral part technology plays in contemporary life, this constitutes a dire danger to the open market, and well deserves DOJ scrutiny.”

Bickley said that the real reason for any investigation is the “battle of mega corporations currently going on. Basically, you have the hyperscalers who are the largest consumers of Nvidia’s GPUs. They’re also trying to develop their own version of GPUs. And no one in the world has been successful at displacing Nvidia in this marketplace. They have close to 90% market share. So Nvidia is pulling levers to drive their revenue and ancillary products and services like the CUDA platform, like their cabling, like their server racks.”

The company’s biggest customers, he said, “are fighting back and saying, ‘we want to do it our way, we need you to adapt, we want the number and quantity of chips we want.’ You have these battles going on.”

According to Bickley, anything related to an antitrust probe is “more hype than substance at this point, because, first, semiconductors are the most volatile sector in the stock market. [Nvidia CEO] Jensen Huang knows full well that this wave that they’re riding at the end of the day will crest, and it will crash, and it will happen violently when it does. They are squeezing everything they can out of the short runway and they are trying to elongate that runway as much as possible.”

As for the DOJ, he said, “if you have reputable names complaining, they are going to take a look at it, but under the Biden administration, the department has been extraordinarily unsuccessful in about every antitrust action they’ve decided to take on. Just because they’re looking at it doesn’t mean they’re going to do anything about it.”

“[By the time] they do something about it, and the time that it takes to resolve, this cycle will have already crested, in my opinion. The money will already be made, the damage will already be done … The reality is that we are talking about an anti-competition inquiry for a product that there is really no competition for.”

That is, he said, “the first thing that comes to my mind. The second is, is they (DOJ) may attack the CUDA platform, which only works for development on Nvidia GPUs. The EU has a practice of saying ‘you can’t do this; you have to make your platform open.’ I think that’s arguable. But why would a platform designed to support a specific piece of hardware have to support other people’s hardware?”

The DOJ could force them to do that, and what Nvidia then might do, said Bickley, is develop an alternative platform that will ultimately be less effective on non-Nvidia products in order to comply.

“And the reality is this, and I will use the Adobe example,” he said. “If I’m a graphic designer, and I go to school, and I’m trained on Photoshop, and it integrates with the other Adobe suite of products you can’t make me use Corel Painter to do what Photoshop does and think I’m not going to go find another job.”

It’s the same thing with developers in the GPU space, he said. “They only want to learn one platform; they don’t have the bandwidth to learn up on multiple platforms. The lock-in is real, and by the time these things [antitrust probes] come around to any sort of fruition, the ecosystem is going to be pretty much set in stone.”

On the heels of Apple rolling out an early preview of iOS 18.1 with its first generative AI (genAI) tools, IDC this week released a report saying nearly three-in-four smartphones will be running AI features in four years.

Apple first announced its plans to add AI features to devices through its Apple Intelligence platform at its developers conference in June. This week, Apple announced AI tools and Siri enhancements that will be exclusive to the iPhone 15 Pro and Pro Max, as well as later iPhones. Apple Intelligence will also be released on Macs running Apple Silicon M-series chips.

Over the next two years, the number of genAI-enabled smartphones is expected to grow quickly, though they’ll likely be limited to flagship devices like the iPhone 15 Pro, Google Pixel 8 Pro, and Samsung Galaxy Z Flip6 — and cost will be an issue.

“Capable chipsets don’t come cheap,” said Anthony Scarsella, IDC’s research director for mobile phones. “Over time, we believe these components will enter the mid-market and more affordable models as competition grows among device manufacturers and AI applications.”

Most high-end mobile processors now include an AI accelerator or neural processing unit (NPU). The Arm processor architecture provides an NPU component for licensees and mobile chips from Intel, Qualcomm, MediaTek, and Samsung have their own NPU implementations, according to Jack Gold, principal analyst with tech industry research firm J. Gold Associates.

“Right now it’s mostly in the high end of devices (e.g., high-end Snapdragon chips), but eventually they will migrate down to the lower end as well,” Gold said. “In the next two to three years, you will be hard pressed to find a mobile processor, except perhaps at the very low end for $100 smartphones, that doesn’t have AI acceleration built in.”

One of the biggest advantages of running genAI on smartphones is that user data never moves off the device — boosting both privacy and security — and locally running apps can benefit from the use of AI accelerations (e.g, audio, video, digital assistance, device management, etc.), according to Gold.

GenAI on mobile devices will automate certain functions, such as generating new messages and emails by choosing different conversational tones such as “friendly,” or “professional” or “concise”, for example. GenAI tools can also be used to pull bulleted points from digital conversations and generate bullet points in a “too long; didn’t read” (TLDR) format, for example.

“Search will always be a top priority and use case for mobile users,” Gold said. “AI-assisted search will not only be for cloud based search, like Google, but also on-device search like, ‘Where did I put that file and picture I took of my dog?’ Beyond that, help with pictures and video will be a major use (and already is) that helps mobile users create better pictures and video.”

For example, the Apple Intelligence platform will allow users to generate images from word prompts. Inputing the word “rooftop” will generate the image of a rooftop scene in a city, which can then be animated or illustrated andsent in a message. Siri, Apple’s voice assistant, will also use Apple Intelligence to better interpret voice commands and help ensure responses are accurate.

Additionally, general concierge-like services will become more widely available as smartphones using genAI learn more about users and their preferences.

GenAI will ‘will completely transform’ how we use phones

“While it is still too early to know all the use cases that will emerge in the coming years, one thing is for sure — genAI will completely transform the way we interact with our smartphones,” said Nabila Popal, senior research director with IDC’s Worldwide Tracker Team.

In its report, IDC said genAI smartphone shipments are forecast to reach 70% of the market by 2028. This year alone, AI-enabled smartphone shipments are expected to grow with more than 360%, representing 234.2 million phones. That represents 19% of the overall smartphone market in 2024.

As phones capable of running genAI features on the device become more popular, the potential for more personalized and proactive AI assistants becomes increasingly likely, IDC said. “This evolution, driven by consumer demand, application developments, and overall industry growth, promises to make the next decade the most exciting period for the smartphone market,” IDC said.

Experts see AI features and tools moving more to the edge — being embedded on smartphones, laptops and IoT devices — because AI computation is done near the user at the edge of the network, close to where the data is located, rather than centrally in a cloud computing facility or private data center.

Expect a migration from the cloud to the edge…

That’s why AI will migrate away from the cloud, according to Gold.

“We expect the largest portion of all AI workloads to migrate to the edge over the next few years, running close to the intended use case,” Gold wrote in a recent research report. “This is both for improved functionality and latency and increased privacy/security. It’s also necessary to offload the need for movement of large amounts of data to centralized computing platforms in the cloud or data center, which is expensive and impacts performance.”

Edge deployments will include running remote cloud instances on localized devices, so remote hyperscaler products should be well positioned for AI’s edge migration. “This presents a significant opportunity for AI expansion and for vendor fulfillment of systems and services,” Gold said.

Tiffany Yeung, an Nividia product marketing manager, wrote in a blog post that since AI algorithms are capable of understanding language, sights, sounds, smells, temperature, faces and other analog forms of unstructured information, “they’re particularly useful in places occupied by end users with real-world problems.”

GenAI smartphones that feature a system-on-a-chip (SoC) are capable of running on-device AI models more quickly and efficiently, according to IDC. And genAI smartphones can leverage NPUs with 30 tera operations per second (TOPS) — or more — using the int-8 data type, (in other words, an 8-bit integer).

…and bigger upgrade cycles

Despite sometimes long mobile refresh cycles and economic unknowns, genAI capabilities on smartphones will drive upgrade cycles and represent a significant opportunity for both hardware makers and app developers alike, IDC said.

“The dramatic growth seen in 2024 will carry into 2025, with shipments of genAI smartphones expected to grow 73.1% year over year. By 2028, IDC forecasts 912 million genAI smartphone shipments, resulting in a 2024-2028 compound annual growth rate (CAGR) of 78.4%.

“The rapid incorporation of genAI in smartphones is unprecedented in mobile history, with market penetration expected to exceed 60% within the first three years,” IDC’s Popal said. “However, the most significant impact of this evolution is anticipated in 2026, when mid-range devices are expected to adopt this technology, making a momentous leap towards the democratization of GenAI.”